INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Fungal growth is the most frequent cause of spoilage in baked goods. In addition to the great economic losses associated with spoilage, another concern is the possibility that mycotoxins could cause public health problems (20).

Fungal contamination of baked goods is influenced by several factors: the type of product (bread or sweet baked goods), ingredients (type of flour and other dry ingredients), leavening sources (chemical, baker's yeast, or sourdough), size and architecture of the bakery, and conditioning and packaging of the products (slicing, wrapping, and materials used for packaging). Since fungal spores are killed during baking, airborne molds contaminate the baked goods during cooling, slicing, wrapping, and storage operations (20).

The most common spoiling fungi isolated from bakery products belong to the genera Penicillium, Aspergillus, Monilia, Mucor, Endomyces, Cladosporium, Fusarium, and Rhizopus (20, 24). Baked goods can be protected from fungal spoilage by destroying any spores which have contaminated the products (e.g., infrared and microwave radiation), using fungal inhibitors such as ethanol and propionic, sorbic, benzoic, and acetic acids and some of their salts, using modified atmospheres or other packaging techniques, and using sourdough with antifungal activity (20, 25).

The use of sourdough in rye and wheat breadmaking, as well as in other sweet baked goods, increases shelf life due to the presence of lactic acid bacteria (11). To date, only a few studies have reported the antifungal activity of lactic acid bacteria, and many of them have considered food processes other than those used for leavened baked goods. The improved microbial shelf life of sourdough baked products was initially attributed to the organic acids produced by lactic acid bacteria (27, 28). By increasing the amount of sourdough used, the shelf life of bread inoculated with conidia of typical bread molds such as Aspergillus niger, Cladosporium herbarum, and Penicillium verrucosum has been extended. The fungistatic effect was attributed to lactic and, especially, acetic acids produced by lactic acid bacteria (25). Further studies have confirmed that the acetic acid concentration was strictly related to the antifungal activity and that other bacterial metabolites also have inhibitory activity (3, 13, 22). It has been shown that the antifungal activity of sourdough lactic acid bacteria varies and is found mainly in obligately heterofermentative Lactobacillus spp. Within this group, Lactobacillus sanfranciscensis CB1 (31) had the largest spectrum of antifungal activity due to the production of a mixture of organic acids (3).

Regarding other food processes, it has been shown that the culture supernatant of a Lactobacillus spp. mixture from a commercial silage inoculum reduces growth and aflatoxin production by Aspergillus flavus subsp. parasiticus (12). The mycostatic activity of a Leuconostoc mesenteroides strain used in cheesemaking has been reported, but the antifungal substances have not been isolated (29). Recently, new antimicrobial compounds in the culture filtrate of Lactobacillus plantarum, which were active against Pantoea agglomerans and also against molds such as Fusarium avenaceum, have been identified (22).

In this work, we studied the antifungal activity of several sourdough lactic acid bacteria and selected L. plantarum 21B because it showed a very wide spectrum of inhibitory activity against fungi isolated from bakery products. Novel antifungal compounds have been identified, their production has been characterized, and the selected strain was used in sourdough bread production.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Microbial species and culture conditions. Eurotium repens IBT18000, Penicillium corylophilum IBT6978, Penicillium roqueforti IBT18687, Endomyces fibuliger IBT605 (belonging to the Culture Collection of the Technical University of Denmark, Lyngby, Denmark), Aspergillus niger FTDC3227, Aspergillus flavus FTDC3226, Eurotium rubrum FTDC3228 (supplied by the Food Technology Department, University of Lleida, Spain), and Aspergillus niger IDM1, Penicillium expansum IDM/FS2, Monilia sitophila IDM/FS5, Fusarium graminearum IDM623, and E. fibuliger IDM3812 (belonging to the Culture Collection of the Institute of Dairy Microbiology, Agriculture Faculty of Perugia, Italy) were used in this study. All these species are commonly isolated from contaminated baked goods (20, 24).

Production of antifungal compounds by lactic acid bacteria. Twenty-four-hour-old cells of lactic acid bacteria were used to inoculate 0.2% (vol/vol) wheat flour hydrolysate (WFH) broth (pH 4.8) produced as reported previously (8). WFH was used as the culture medium to select for antifungal activity because its main composition is similar to that of wheat flour. Except for the L. fermentum and L. acidophilus strains, which were incubated at 37°C, all the other strains were incubated at 30°C for 72 h. After incubation, cells were harvested by centrifugation (9,000 × g for 10 min at 4°C), and the supernatants were filter sterilized and freeze-dried. After freeze-drying, the bacterial culture filtrates (BCFs) were concentrated 10-fold with respect to their initial volume and used for the conidial germination assay.

Conidial germination assay. The antifungal activity of the BCFs was determined using the conidial germination assay described previously (18), with some modifications. Ten microliters of a conidial suspension in 0.05% (vol/vol) Triton containing about 10³ conidia of the fungal species per ml was added to 190 µl of (i) a solution of 10-fold-concentrated BCFs (500 mg of dry matter per ml, pH 3.6 to 3.8), (ii) a solution of 10-fold-concentrated WFH (500 mg of dry matter per ml, pH 4.8), or (iii) solutions of WFH containing calcium propionate and sodium benzoate at concentrations of 0.003, 0.03, and 0.3% (wt/vol) and incubated at 25°C for 30 min. When E. fibuliger strains were assayed as indicators, a suspension of 10³ spores per ml was used. Chemicals were from Aldrich Chemical Company, Milwaukee, Wis. Aliquots (50 µl) of each mixture were then spread onto the surface of three petri dishes (60-mm diameter) containing 5 ml of PDA. Plates were incubated at 26 or 30°C for 24 h. After incubation, the number of germinated conidia was determined by stereoscopic observations. The assay was repeated three times.

Isolation and assay of the antifungal compounds produced by L. plantarum 21B. Aliquots (20 ml) of the BCF of L. plantarum 21B were initially extracted (1:1, vol/vol) with an eluotropic series of organic solvents, namely n-hexane, chloroform, ethyl acetate (pH 2.0, 3.6, and 10.0), and n-butanol. Ethyl acetate gave the highest recovery of the antifungal activity. BCF (200 ml, pH 3.6) was extracted four times with 200 ml of ethyl acetate, and the combined organic extract of culture filtrate was dried (Na₂SO₄) and evaporated under reduced pressure to give a crude residue of ca. 338 mg. Thin-layer chromatography (TLC) analysis of the residue was performed on silica gel plates (Merck, Kieselgel, Germany; 60 F₂₅₄, 0.25 mm). The spots were visualized by exposing the plates to UV radiation and spraying with 10% H₂SO₄ in CH₃OH and then with 5% phosphomolybdic acid in CH₃OH, followed by heating at 110°C for 10 min. The residue (100 mg) was partially purified by preparative TLC (Merck; 60 F₂₅₄, 0.5 mm) to give three fractions, A, B, and C (5.2, 4.5, and 44.0 mg, respectively). The extract and fractions were dissolved with CHCl₃-methanol (MeOH) (1:1, vol/vol), and inhibitory activity against E. fibuliger IBT605 and P. roqueforti IBT18687 was determined by the antifungal disk assay. Fifty-microliter aliquots containing crude extract or fractions (corresponding to 20 ml of BCF for crude extract and 60, 60, and 30 ml of BCF for fractions A, B, and C, respectively) or 50-µl aliquots of CHCl₃-MeOH (1:1) were added to 6-mm sterile disks (sterile blanks; Difco), allowed to dry, and placed onto PDA. Plates were overlaid with soft agar (2 ml, 0.7%) containing about 10³ cells per ml of test fungi. After incubation at 25°C for 24 h, the inhibition areas on and around the disks were recorded.

Identification of the antifungal compounds by GC/MS. For gas chromatography/mass spectrometry (GC/MS), samples of ethyl acetate extract (7 mg) of the L. plantarum 21B culture filtrate and fractions (0.5 mg) A, B, and C obtained from its preparative TLC were esterified by dissolving in CH₃OH and treated with an ether-diazomethane solution for 30 min at 0°C with stirring. The mixtures were evaporated with an N₂ stream. A QP5000 GC/MS instrument (Shimadzu, Kyoto, Japan), equipped with a nonpolar capillary column MDN-5 (30 m by 0.25 mm inner diameter; film thickness, 0.25 µm; Supelco, Belleforte, Pa.), was used for the analysis. The GC oven was held at 140°C for 1 min and then increased to 240°C at 4°C per min for 5 min. Helium was used as the carrier gas with a flow of 30 ml/min. The identification of the compounds was based on 90% similarity between the MS spectra of unknown and reference compounds in an MS spectra library.

Sourdough fermentation with L. plantarum 21B. The wheat flour contained moisture (12.8%), protein (N × 5.70), 10.6% dry matter (d.m.), fat (1.79% d.m.), and ash (0.60% d.m.). Wheat flour (250 g), 110 ml of tap water, and 40 ml of cellular suspension, containing 10⁷ CFU of S. cerevisiae IDM141 (Culture Collection of the Institute of Dairy Microbiology, Agriculture Faculty of Perugia, Italy) per ml and L. plantarum 21B (10⁹ CFU ml¹), S. cerevisiae IDM141 and L. brevis 1D (10⁹ CFU ml¹), or S. cerevisiae IDM141 alone were mixed to produce 400 g of dough (dough yield = 160) with a continuous-high-speed mixer (60 × g; optimal dough mixing time, 5 min) (Chopin & Co., Boulogne, Seine, France). Doughs were individually placed in aluminum pans (25 cm by 10 cm by 8 cm high) and incubated at 30°C for 150 min. After fermentation, the three sourdoughs were baked in a batch oven (Mondial Forni, Verona, Italy) at 220°C for 30 min. Sourdough breads were then cooled at room temperature for 90 min, sliced (10 cm by 1 cm by 9 cm high), and 2 ml of conidial suspension of A. niger FTDC3227, prepared as previously reported and containing about 10⁴ conidia per ml, was spread by nebulization on the slice surface. The slices were then packed in polyethylene bags, 95-µm thick (Tillmans S.p.a, Milan, Italy), and stored at 20°C for 7 days.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Preliminary screening. After 72 h of incubation in WFH broth, all the lactic acid bacteria assayed reached a cell number of ca. 10⁹ CFU ml¹ and all the BCFs used in the conidial germination assay had a pH of between 3.6 and 4.0.

Spectrum of antifungal activity of selected lactic acid bacteria. L. plantarum 20B and 21B, L. alimentarius 5Q, L. lactis subsp. lactis 11M, and L. citreum 10M were selected because of their greater inhibitory spectra, and the fungicidal activity was further characterized by using eight other fungal species or strains as indicators (Table 1). F. graminearum was omitted because it was inhibited by all 25 strains of lactic acid bacteria assayed and different strains were used for the species considered as indicators in the preliminary screening. The activity was compared with antifungal chemicals such as calcium propionate and sodium benzoate.

1

1

1

1

Characterization of the antifungal activity of L. plantarum 21B. When cultivated in WFH broth, L. plantarum 21B reached the stationary phase (3.0 × 10⁹ CFU ml¹) after 48 h, then cell viability decreased dramatically after 7 days of incubation at 30°C (Fig. 1). The production of lactic acid followed a similar trend, reaching a maximum of 8.8 mmol liter¹ after 48 h, while acetic acid and ethanol productions were very limited. Fungicidal activity against E. fibuliger IBT605 was detected when the cell number was the highest; it increased slightly and then remained constant over 10 days of incubation. When the pH of BCF from L. plantarum 21B was changed from 5.0 to 7.0, the inhibition of E. fibuliger IBT605 decreased markedly, but by readjusting the pH of BCF to the initial value of 3.7, the antifungal activity was restored (data not shown). In sourdough baked goods, pH values of 3.7 to 4.0 are very common (11). The antifungal activity was heat stable to 100°C for 15 min (data not shown).

View larger version (18K): [in this window] [in a new window]   FIG. 1.   Kinetics of growth, fermentation end products (lactic and acetic acids and ethanol), and antifungal activity of Lactobacillus plantarum 21B. Antifungal activity is expressed as percent inhibition of spore germination of Endomyces fibuliger IBT605. Each value is the mean from three replicates ± standard deviation.

Isolation and identification of antifungal compounds produced by L. plantarum 21B. First, BCF from L. plantarum 21B was exhaustively extracted by using ethyl acetate at pH 3.6. This organic extract still showed 100% inhibitory activity against E. fibuliger IBT605. The crude extract was then fractionated by preparative silica gel TLC, and three fractions (A, B, and C) were recovered. The crude extract (5.5 mg/disk) and fractions A, B, and C (0.8, 0.7, and 2.5 mg/disk, respectively) were assayed by the antifungal disk assay using E. fibuliger IBT605 as the indicator. Only the crude extract and fraction C, containing the more polar compounds, caused a 3-mm inhibition halo around the disks. The same results were found when the crude extract and fractions (concentration of 16, 2.4, 2.1, and 7.5 mg/disk) were tested against P. roqueforti IBT18687. The solvent CHCl₃-MeOH did not inhibit any fungal species. The ethyl acetate extract and all three fractions were analyzed by GC/MS after converting the organic acids to the corresponding methyl esters by reaction with diazomethane. The compounds were identified by comparing their electron impact (EI)-MS spectra with those recorded in the mass spectrum library; identification was considered reliable only if there was >90% similarity with the reference spectrum. The compounds identified in the inhibitory fraction C corresponded to phenyllactic acid, p-hydroxyphenyllactic acid, and palmitic acid in the ratio of 4.2:1.5:1.0 (Fig. 2b). These compounds were also included in the GC/MS profile of the crude extract (Fig. 2a), which was more complex due to the presence of other substances subsequently fractionated by preparative TLC and contained in the other two noninhibitory fractions (A and B).

View larger version (22K): [in this window] [in a new window]   FIG. 2.   GC profiles of ethyl acetate extract (a) of culture filtrate from Lactobacillus plantarum 21B and (b) of active fraction C obtained from its purification. Metabolites were identified by GC/MS analysis and compared with standard samples.

Antifungal activity of the identified organic acids. The lowest inhibitory dose of phenyllactic and p-hydroxyphenyllactic acids against E. fibuliger IBT605 (ca. 1.5-mm inhibition halos around the disks) corresponded to 2.5 and 6.6 mg/disk, respectively. Phenyllactic acid was also assayed by the conidial germination assay (concentration of 14 mg/ml of WFH), and the activity was confirmed to be as fungicidal, as were those of BCF from L. plantarum 21B (data not shown). Palmitic acid at the highest concentration tested (10 mg/disk) did not inhibit the indicator fungus. When the inhibitory activity of organic acids was tested against P. roqueforti IBT18687, only phenyllactic acid (8.3 mg/disk) caused an inhibition halo of ca. 1 mm around the disk. Palmitic acid and p-hydroxyphenyllactic (10 and 16 mg/disk, respectively) were not inhibitory. When phenyllactic, p-hydroxyphenyllactic, and palmitic acids were used in a mixture according to the ratio found in fraction C (8.3, 3, and 2 mg/disk, respectively), inhibition of E. fibuliger IBT605 (halo of ca. 3 mm) did not increase over that with phenyllactic acid alone (8.3 mg/disk).

Sourdough fermentation with L. plantarum 21B. Breads were produced using associations of L. plantarum 21B and S. cerevisiae IDM141, L. brevis 1D and S. cerevisiae IDM141, and S. cerevisiae IDM141 alone. L. brevis 1D is a sourdough strain which did not show appreciable antifungal activity during the preliminary screening but produced about the same amount of lactic acid as L. plantarum 21B in addition to acetic acid. Both breads started with lactic acid bacteria had low pH values (4.4 to 4.6), while those started with the yeast alone had a pH of 5.7. After baking, the breads were sliced, and a conidial suspension of A. niger FTDC3227 was spread by nebulization on the slice surface. The slices were packed in polyethylene bags to maintain constant moisture and then stored at 20°C for 7 days. Figure 3 shows how mold growth occurred mostly after 2 days in the breads started with S. cerevisiae IDM141 alone (the same was found for the association with S. cerevisiae IDM141 and L. brevis 1D), while the selected L. plantarum 21B delayed mold contamination until after 7 days of storage.

View larger version (85K): [in this window] [in a new window]   FIG. 3.   Growth of Aspergillus niger FTDC3227 in bread started with S. cerevisiae 141 and Lactobacillus plantarum 21B after 7 days of storage (A) and in bread started with Saccharomyces cerevisiae 141 alone after 2 days of storage (B).

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Lactic acid bacteria greatly influence the sensory, textural, nutritional, and shelf-life characteristics of sourdough baked goods, especially breads (11). Antifungal activity has to be considered an important tool for selecting sourdough lactic acid bacteria because fungal, rather than bacterial, spoilage is the main cause of substantial economic loss in the baking industry and may also cause public health problems due to the production of mycotoxins (19, 20, 28).

In this study, several sourdough lactic acid bacteria were screened, and novel antifungal compounds from selected Lactobacillus plantarum 21B were purified and characterized. L. plantarum 21B showed a very broad spectrum of activity and inhibited Eurotium repens IBT18000, E. rubrum FTDC3228, Penicillium corylophilum IBT6978, P. roqueforti IBT18687, P. expansum IDM/FS2, Endomyces fibuliger IBT605 and IDM3812, Aspergillus niger FTDC3227 and IDM1, A. flavus FTDC3226, Monilia sitophila IDM/FS5, and Fusarium graminearum IDM623. These fungi represent almost all the species most commonly isolated from contaminated baked goods (20). Phenyllactic acid, its corresponding 4-hydroxy derivative (p-hydroxyphenyllactic acid) and palmitic acid were identified by GC/MS analysis in the active fraction of the BCF from L. plantarum 21B grown in WFH broth. When the same commercial compounds were used individually, only phenyllactic and p-hydroxyphenyllactic acids showed antifungal activities. A cooperative action has often been reported for antimicrobials produced by microorganisms, because in mixtures such compounds may interact with each other as well as with the test organisms (3, 5, 22). In this study, no synergistic effect of the mixture was found, showing that phenyllactic acid played a key role in inhibiting fungal growth.

Strain L. plantarum 20B, which had broad antimold activity, was also selected. The preliminary purification and characterization of the antifungal compounds produced by strain 20B also led to the identification of phenyllactic and p-hydroxyphenyllactic acids (data not shown).

To our knowledge, this is the first report showing the production of phenyllactic acid and p-hydroxyphenyllactic acid by lactic acid bacteria. Both compounds are involved in phenylalanine metabolism (26). To avoid intracellular accumulation of phenylalanine, this amino acid may be hydroxylated to tyrosine or transaminated to phenylpyruvic acid, which is further metabolized to phenylactic and p-hydroxyphenyllactic acids. Experiments conducted in Candida species with [¹⁴C]phenylalanine confirmed that 2-phenyllactic acid was synthesized from L-phenylalanine (21). Based on these findings, it may be supposed that phenyllactic acid was excreted in copious amounts during growth of L. plantarum 21B in WFH to avoid intracellular accumulation of phenylalanine. Neither phenylalanine or tyrosine is an essential or stimulatory amino acid for most of the sourdough L. plantarum strains (9). D-3-Phenyllactic acid was purified and characterized from Geotrichum candidum, a yeast-like fungus used in the maturation of several cheeses (5). A patent application used mutants of Brevibacterium lactofermentum to produce D-3-phenyllactic acid (16). In an extensive study on the interactions among cheese microflora, it was found that G. candidum produced D-3-phenyllactic acid which inhibited fungi as well as gram-negative and gram-positive bacteria (14). Other authors (4) used D-3-phenyllactic acid against Listeria monocytogenes grown in culture medium and in milk. They found a bactericidal effect independent of the physiological state of the pathogen and a reduction in the milk population of ca. 4.6 log cycles. A phenyllactic acid concentration ranging from 10 to 20 mg ml¹ was needed to inhibit L. monocytogenes, Staphylococcus aureus, and Enterococcus spp. in the agar well diffusion assay (5, 6). Although different biological tests were applied and the target organism was a fungus, in our conditions, 2.5 mg (which corresponded to 50 mg ml¹) of phenyllactic acid per disk was inhibitory to all the fungi tested except P. roqueforti IBT18687 and P. corylophilum IBT6978 (166 mg ml¹).

The activity of the BCF of L. plantarum 21B was compared to those of chemicals widely used as preservatives in foods. Conidial germination assays with 0.3 to 3 mg of calcium propionate ml¹ had practically no antifungal activity. Sodium benzoate showed antifungal activity only at 3 mg ml¹ which varied slightly in the inhibitory spectrum compared to that of L. plantarum 21B. Directives of the European Community (1) permit the use of a maximum of 3 mg of calcium propionate ml¹ for packaged sliced breads and rye breads and recommend potassium sorbate at a concentration of 2 mg ml¹, which was not inhibitory against E. fibuliger IBT605, A. niger FTDC3227, or P. roqueforti IBT18687 (data not shown). Potassium sorbate is rarely used because of the secondary effects on bread volume.

The potential of antimicrobials other than bacteriocins produced by lactic acid bacteria is currently being exploited due to the very broad spectrum of activity. After the structural characterization of reuterin by Lactobacillus reuteri (30), pyroglutamic acid, produced by Lactobacillus casei subsp. casei, has also been introduced as an antimicrobial agent (15). Recently, antimicrobial compounds have been identified in the culture filtrate of another strain of L. plantarum (22). They corresponded to benzoic acid, 5-methyl-2,4-imidazolidinedione, tetrahydro-4-hydroxy-4-methyl-2H-pyran-2-one, and 3-(2-methylpropyl)-2,5-piperazinedione and were inhibitory to Pantoea agglomerans and also inhibited the fungus Fusarium avenaceum to some extent.

For a long time, the improved shelf life of sourdough baked products was attributed to the lactic and acetic acids produced by lactic acid bacteria (27, 28). Further studies (3, 13, 22) have shown that lactic acid is not inhibitory to fungi, while the acetic acid concentration seems to be more strictly related to the antifungal activity (25). The acetic acid concentration in the dough may be increased by adding fructose, which is used as an external electron acceptor by heterofermentative lactic acid bacteria, which consequently increases the growth yield and acetic acid production (10). Very few studies have focused on the antifungal activity of sourdough lactic acid bacteria. A previous screening of sourdough lactic acid bacteria showed that L. sanfranciscensis CB1 (obligately heterofermentative strain) produced a mixture of acetic, caproic, formic, butyric, and n-valeric acids which synergistically inhibited species of Fusarium, Penicillium, Aspergillus, and Monilia. Caproic acid played a key role in inhibiting mold growth (3). It was also shown that sourdough lactic acid bacteria do not have the same antifungal potentialities, which do not depend on the production of lactic and acetic acids. The L. plantarum strains (facultatively heterofermentative) of this study have a very large spectrum of activity based on antifungal compounds which are different from those identified previously (3). It must be emphasized that the association of L. sanfranciscensis and L. plantarum is very often identified in sourdoughs (11).

The antifungal activity of L. plantarum 21B was also found in sourdough bread. Compared to breads started with S. cerevisiae 141 alone or in association with L. brevis 1D, the sourdough bread which used L. plantarum 21B in association with S. cerevisiae 141 delayed fungal contamination until after 7 days of storage at room temperature. When L. plantarum 21B was grown in a culture medium richer in nutrients, such as the WFH, the inhibitory activity was found to be ca. 90% of that in the 10-fold-concentrated BCF, showing that the antifungal activity may be increased depending on the type of flour used.

To be effective and competitive, the food industry must respond to consumer demands, and recent trends have included the desire for high-quality foods that are not extremely processed and that do not contain chemical preservatives. Because the antimicrobial compounds produced by lactic acid bacteria are considered natural preservatives, the use of L. plantarum 21B to decrease the fungal contamination of sourdough baked products has interesting potential applications.